This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Background Anesthetic agents modulate a number of voltage gated and ligand gated ion channels, but there does not seem to be specific channels that are the site of action for all anesthetics. Even when the site of action of a specific anesthetic is known, the process whereby anesthetic modulation of ion channels leads to the anesthetic state has yet to be established. The overall goal of this research is to use large scale computational models to elucidate the integrated systems level response of model neurons to anesthetics. The data generated from computing resources requested in this application will be used to support an application to NIH for additional resources to continue this work. To appreciate the importance of building a computational bridge between anesthetic action at the receptor level and the systems level, it is necessary to realize the fundamentally different responses to anesthetics in each domain. A single anesthetic can modulate the activity of one or more voltage gated and/or ligand gated ion channels. Typically, the concentration effect curves for a given channel are relatively shallow with midpoints occurring at anesthetic concentrations that are well above those used clinically. This graded modulation of ion channel behavior contrasts the abrupt changes that anesthetics induce at the systems level. For clinically useful anesthetic concentrations the brain is far from quiescent, as revealed by electroencephalographic and functional imaging studies. Over a population of subjects, the onset of the anesthetic state is relatively abrupt as anesthetic concentration is increased, and the concentration at which this occurs is considerably below the midpoint of the concentration effect curve for the putative ion channels. The systems level response to anesthetics is additionally complex because it encompasses a number of distinct features that emerge at distinct anesthetic concentrations. Minimally these include amnesia, loss of consciousness, blockade of painful stimuli, and immobility. Importantly, these effects can be produced by anesthetics that target entirely different sets of receptors. The qualitatively disparate behaviors described demonstrate the need for approaches linking anesthetic action at the receptor level with large scale systems level behavior. Although there is no specific network behavior that is definitively linked to general anesthesia, there is a growing appreciation that at least some aspects of general anesthesia are linked to the ability of the brain to generate coherent oscillations. These oscillations are thought to originate in the thalamic circuitry and the synchronization of these oscillations are dependent on the interaction established between the thalamus and the cortex. It is reasonable to hypothesize that these effects on the thalamus and cortex could be an important component of the anesthetic state since this region has already been shown to be closely associated with both sleep and consciousness. Furthermore, our preliminary results have demonstrated the ability of anesthetics to both synchronize and slow oscillations in a model of the thalamic relay and reticular nucleus neurons. The possible impact of these alterations on the rest of the thalamus and its interaction with the cortex are important considerations which have yet to be examined. To date, we have examined the effects of a variety of anesthetics on single cell and small network models of the reticular nucleus of the thalamus, thalamocortical neurons, hippocampal neurons, a fast spiking interneuron network, and Aplysia. We now seek to expand this effort to incorporate neurons which are more realistic with respect to types of ion channels and morphology. Preliminary results in small networks (2 cells of each neuron type) incorporating 4 different types of neurons, pyramidal neurons (PY) and interneurons (IN) in the cortex and thalamic relay (TC) and reticular nucleus (RE) neurons in the thalamus, have shown that the feedback between the thalamic neurons and the cortical neurons are important in understanding the behavior of the neurons under anesthetic effects. Being able to create large complex networks (100 cells of each neuron type) is essential to discerning the differences in anesthetics on the overall system level behavior and at the cellular level. Goals The following represent our immediate research goals with the requested support from PSC: 1. Continue our development of detailed pharmacologic models of anesthetic modulation of realistically complex biophysical neural models, where we now extend this work to the neurons of the thalamus and cortex. 2. Examine the ability of these networks to synchronize their activity as a function of the extent of pharmacologic intervention. We have already constructed several models of individual thalamic neurons and networks. The models have thus far been rather limited in complexity. It is clear that a greater level of model detail will permit the study of anesthetic effects that have heretofore been oversimplified. Increasing the level of detail in both ion channels and morphology is critical in assessing the integrative aspects of anesthetic action, as small and subtle changes in cell behavior can have large effects when incorporated into a larger network. A model endowed with sufficient detail in ion channel population, cell morphology and network interaction can be constructed from published models of thalamic cells and circuits. Each cell consists of multiple compartments representative of their morphology, contains both voltage gated and calcium controlled ion channels, sophisticated calcium concentration handling, and connects to the network through inhibitory GABAA channels. Supercomputer Use The goal in this phase of the research is to develop the network described above in a parallel environment using Parallel Genesis, a simulation environment already supported by the PSC. In the process of developing this network, preliminary study of the individual neuron and of the network in the presence of anesthetics will be conducted. Resources Needed Based on simulations of a network of 100 single compartment cells developed in Mathematica and run on a single 3.0 GHz processor, we estimate we will need 10,000 service units on the Terascale Computing System (TCS) to accomplish these goals. The use of implicit numerical methods in solving the equations permits a step size of .01 ms. With this step size, a 100 single compartment cell network requires approximately 20,000 cpu seconds to run for 10,000 simulation milliseconds (or 1*10^6 steps). As the TCS consists of 1 GHz processors, we use the approximation that 10,000 simulation seconds will require 20,000 cpu sec*(3 Ghz/1 GHz)/(3,600 sec./hr.) = 16.7 Service Units (S.U.). This is the standard duration for our simulations. Further development is intended to increase the complexity of the cells from a single compartment model to a three compartment model, as well as including 4 neuron types in the network. The following represents how we plan to use these resources: Adapting Scripts to Run Under PGENESIS 1,000 S.U. for development of existing models in parallel Genesis Simulation of Anesthetics in Thalamic Network 9,000 S.U. for 30 runs of 100 TC + 100 RE + 100 PE + 100 IN Three Compartment Cells 9,000 = (16.7 S.U.)*(4 Types of Cells)*(3 compartment scaling)*(30 runs)* (1.5 parallel scaling factor) = 1,000 + 9,000 = 10,000 S.U.